Geoscience Reference
In-Depth Information
There are a number of reasons why the consistency between the results of
the detailed study of individual examples and the statistical scrutiny of a rela-
tively larger number of observations is valuable. One reason is that observable
data at a particular time are the result of medium to long-term processes, such
as those studied earlier in this topic. The observable relationships between
any phenomena-in this case economic geography and the structure of politi-
cal representation-are the legacy of history. This means that they are short-
term manifestations of the long-term interplay between these two variables of
interest.
To the extent that the examples analyzed represent the dominant empirical
regularities within the universe of interest, the different empirical approaches
to the relationship between economic geography, political representation, and
fiscal structures ought to yield broadly consistent findings. Should this be the
case, the central claims in this topic can be more widely generalized, even if
only within the small number of unions for which there is available data.
As the observable data is historical, scholars must work with small, non-
random segments of a longer term, evolving relationship between institutions,
geography, and outcomes. Two major implications follow: the first concerns
causality, the second concerns data availability and the strategy behind the
organization of the chapter.
Regarding causality, let me briefly return to the more extensive discussion
in Chapter 3 . The generation process behind the available data makes causal
identification a challenging task. I will make use of an instrumental variable
approach to minimize the problems associated with the fact that fiscal struc-
tures, inequality, and representation are endogenous over time. However, the
use of instrumental variables is a source of growing skepticism in that their
proposed solution appears optimal in theory but is rarely, if ever, to be found
in the real world. With the exception of some structural geographic features,
the proposed instruments seldom meet restriction assumptions. This is because
they consist of variables that are part of the same long-term process, and these
are bound to be correlated in some form or another.
In light of these concerns, this chapter makes a limited yet crucial contribu-
tion to the overall project. Rather than being the ultimate proof of a particular
causal effect, I see the analyses below as a correlational validation of a set
of expectations about the nature of the association between economic geog-
raphy, representation, and redistribution. In other words, if the central claim
of the argument remains firm and the case study findings broadly apply, the
examination of any medium to large subset of political unions ought to yield
a particular set of systematic patterns. Assessing this condition is the ultimate
goal of this chapter.
In this endeavor, a common trade-off between data quality and the spatial
and temporal scope of the study is faced. As we shall see below, this trade-
off is particularly acute when the necessary data involves both the nature and
organization of fiscal structures and the distribution of income. To overcome
Search WWH ::




Custom Search